runRABIT: runRABIT
In caleblareau/RABIT: Regression Analysis with Background InTegration

Description Usage Arguments Details Value

Performs regression analysis

1 2	runRABIT(x, y, b = matrix(0), c = matrix(0), f = 0.05, t = TRUE, s = FALSE, r = TRUE)

`x`	Input variable matrix. Each column represents a variable to be selected. REQUIRED.
`y`	Response matrix. Each column represents a response vector and Rabit will run variable selection on each response separately. REQUIRED.
`b`	Background factors. Each column represents a confounding factor to be controlled. Rabit will exclude these effects from the linear model. Default = NULL
`c`	Background factors for each column of Y. Different from b which inputs background factors shared across all columns of Y, this input is specific for each column of Y. Rabit will search for shared column names between Y and input background factors here and control these factors. default = NULL
`f`	FDR threshold. Range (0,1]. Rabit estimates the statistical significance of each variable in X and only keeps significant ones in later stepwise forward regression. Set to 1 to skip this step. default = 0.05
`t`	Transform Y to normal distribution ( = TRUE) or not ( = FALSE). Transforming response Y to normal distribution increases the statistical power of linear regression t-test. However, if Y is very different from the normal distribution (e.g. binary outcome), then this methodology should not be used and logistic regression should be used in this example. default = TRUE
`s`	Select one best variable in X for each category ( = TRUE) or not ( = FALSE). When several variables in X come from the same category (e.g. Transcription factors may have multiple ChIP-Seq profiles available), multiple names appended together with a . (e.g. CatA.V1, CatA.V2 ...) Rabit will chose the best one based on the maximum t-value and these variableswill be returned. default = FALSE
`r`	Run forward stepwise selection ( = TRUE) or not ( = FALSE). If set to FALSE, Rabit will only calculate the statistical significance of each variable without forward stepwise selection. default = TRUE.